In recent years researchers and policy-makers have become increasingly concerned about the paucity of information on women's health. One area that is especially important is the health effects of work for women, particularly since women continue to move into the labor force in increasing numbers. Unfortunately, estimating the health effects of employment have proved to be difficult, primarily because much of the complexity of women's worklives has yet to be examined. The primary objective of the proposed research is to develop more realistic models of women's worklives and to understand the causal relationship between work and the physical and emotional health of mid- life women. The study is divided into two stages. The first stage addresses the effects of health, working conditions and fringe benefits, unpaid work, and family roles on subsequent patterns of paid and unpaid work. The second stage builds on the first to examine the effects of baseline characteristics and work patterns on health. Comparisons will be made between Black and White women throughout the study. The National Longitudinal Survey of Mature Women (NLS) will be used to address these issues. The NLS is a nationally representative sample examining the labor market experience of over 5,000 women. The original sample was selected in 1967 and subjects have been re-interviewed approximately every two years. Baseline measures of health, working conditions, unpaid work and family roles will be obtained from the 1982 survey, when the women were-between 45 and 59 years old. Patterns of paid and unpaid work will be examined from 1982 to 1989, and physical and emotional health will be assessed in 1989. As of 1989, approximately 3100 women remained in the study, ranging from 52 to 66 years of age. Several dependent variables are proposed and analytic methods include logistic regression, Poisson regression for count data, and simultaneous equation modeling. Assessment and correction for any sample selection bias will be addressed throughout the proposed study.